How does the interpretability of vision-language models differ from traditional computer vision models?Answer not yet generated.